Check That AI Can Reach Your Site
What you'll be able to do
- Find out in two minutes whether AI can actually read your website
- Judge the answer properly, so a confident-sounding summary doesn't fool you
- Know the three questions to send your web person if the answer is no
▶ Video coming soon
Walkthrough for this lesson
KEY TAKEAWAYS
- AI engines read your website using automated visitors called bots, and your site can turn them away by accident
- The whole test is one prompt: ask an AI to read your site and describe your business back to you
- Check what it says against your actual page. If the summary is vague or wrong, treat that as a failure even when the AI sounds sure of itself
- If AI can't get in, the block is in one of three places, and your web person can check all three in minutes
- While you're in there, register your site with Bing, because ChatGPT and Copilot lean on Microsoft's index
Why this matters
When ChatGPT names a business, it read that business's website first. AI reads websites using automated visitors called bots, and your website can be set up to turn those visitors away, almost always by accident. A security plugin, a hosting default, or a checkbox somewhere labeled “block AI scrapers” is enough to do it.
If that's happening to you, nothing else in this course can help. It's the locked front door problem: the shop can be beautiful inside and it makes no difference. Two minutes rules it out.
Do this now
- •Open a fresh chat in ChatGPT, Perplexity, or Gemini. Fresh matters, or it may answer from something you told it earlier instead of from your site.
- •Run the Can AI Read Your Website prompt below. Your web address is the only thing it needs from you.
- •Read the bolded line at the top. That's your answer.
- •Read the three-sentence summary underneath it and check it against your real page.
How to read what comes back
The summary is the real test, not the yes or no. A chatbot that cannot reach your site will sometimes produce a confident description of your business anyway, assembled out of your business name and your industry. That is why the prompt makes it quote three phrases from your page word for word. Go and find those phrases on your site. If they are there, it really read you.
A pass looks like specifics that could only have come from your pages: your actual services, your city, a sentence you remember writing. A fail is the AI saying it can't open the site, or handing back something smooth and general that would fit any competitor in your category. Treat vague as failed, because it usually means the AI is guessing.
One honest limit. This tells you whether AI can read your site when a person asks it to. It is not a cast-iron guarantee about the separate bot that builds the index behind “who does this in my city.” But a failure here is always real, and the same handful of settings sits behind both, so a pass means you're clear to move on and a fail means you have something to fix.
If AI can't reach you
This part isn't yours to fix, and you don't need to understand it. A block lives in one of three places, and only whoever manages your website can see all three:
- •A file called robots.txt, which works like a guest list for automated visitors.
- •A security or SEO plugin. On WordPress especially, these ship with settings like “block AI scrapers” that get switched on without anyone remembering doing it.
- •Your CDN or firewall, the service sitting in front of your website. Cloudflare, one of the most widely used, has a one-click setting that blocks AI crawlers, and in 2025 it began switching that on by default for newly added sites.
So email your web developer, agency, or host and ask them three things. Is our robots.txt allowing AI search crawlers? Is any plugin blocking AI bots? Is AI crawler blocking switched on at Cloudflare, or whatever service sits in front of the site? Ask them to reply saying which ones they checked and what they changed. It is a five-minute job for them, and once they confirm, re-run the prompt and watch the answer flip.
The one decision that's yours to make
If your web person asks which bots you actually want, there are two kinds, and only one of them is a real choice:
- •Bots that help customers find you. These are how you get mentioned and linked in AI answers, so allow them. Blocking one takes you out of those answers entirely. (Big news publishers often block them on purpose, because they want to charge AI companies for their articles. You're trying to get found, so that logic isn't yours.)
- •Bots that collect text to train future AI models. This is a genuine business decision, not a visibility one. Allowing them means your writing may shape how AI talks about your industry down the road. Blocking them is a perfectly fair stance if you'd rather not contribute. Either way, it does not change whether you show up in ChatGPT tomorrow.
Two more things while you're here
- •Register with Bing. Most owners haven't thought about Bing in years, but ChatGPT and Microsoft Copilot both lean on Microsoft's index, so if Bing can't see you, a big slice of AI search can't either. Bing Webmaster Tools is free, takes a few minutes, and lets you submit your sitemap right there. Do the same in Google Search Console if you never have.
- •Ignore llms.txt. If anyone offers to add a file called llms.txt to make your site AI-friendly, say no. No major AI engine reads it, Google has said it doesn't support it and has no plans to, and nobody should be billing you for it.
Put it into practice
Run the prompt on your homepage and check the three quoted phrases against your real page. If it read you accurately, write down today's date and move on. If it couldn't, send the three questions to whoever manages your site today.
PROMPTS FOR THIS LESSON
Copy these into your AI tool. They auto-fill with your business profile. Browse the full library.
Can AI Read Your Website?
Give an AI your web address and find out in one step whether it can actually read your site, with a summary of your business as the proof.
What's next
The door is open. Next you'll find out whether what's behind it is worth quoting.